# linkedin-skills

A collection of LinkedIn skills for AI agents (Claude Code, Codex, Cursor, Windsurf), powered by [LinkedIn CLI](https://github.com/Linked-API/linkedin-cli) (`@linkedapi/linkedin-cli`) and [Linked API](https://linkedapi.io).

## Install

One command — detects your installed agents, asks what to install and where, and sets up the prerequisites for you:

```bash npx @linkedapi/skills ```

Or just hand this to your AI agent:

```text Read https://linkedapi.io/skills/install.md and follow it. ```

For a single skill, hand over its own runbook instead — e.g. `https://linkedapi.io/skills/linkedin/install.md` or `https://linkedapi.io/skills/linkedin-growth/install.md`.

### Non-interactive (agents / CI)

```bash # Inspect the environment first (machine-readable) npx @linkedapi/skills detect --json

# Install one skill in one command npx @linkedapi/skills add linkedin --yes

# Install specific skills into specific agents npx @linkedapi/skills add linkedin linkedin-growth \ --agent claude-code --agent codex --scope project --yes --json ```

Other commands: `list`, `update`, `remove`, `doctor` (all support `--json`). See `npx @linkedapi/skills --help`.

#### Exit codes & readiness

`add` separates two things so a caller never mistakes "needs setup" for "install failed" (the agent-facing install runbook at `linkedapi.io/skills/install.md` spells this out for AI agents):

- **Exit code** reflects only whether the install itself succeeded (files placed + dependencies installed). `0` = installed. `1` = a real install failure. `2` = bad arguments. - **Readiness** is reported in the JSON, not the exit code. Each skill carries `ok` (installed) and `ready` (fully configured and usable now). When a skill installed fine but still needs setup — e.g. LinkedIn tokens not connected yet — you get **exit 0** with `ready: false` and a `pending` list describing what's left. Always parse the JSON; do not treat a readiness gap as a failure.

```jsonc { "success": tr